all_results.json

{
    "epoch": 0.9880609304240429,
    "eval_logits/chosen": 9.35390567779541,
    "eval_logits/rejected": 10.18101692199707,
    "eval_logps/chosen": -0.37907084822654724,
    "eval_logps/rejected": -0.3889653980731964,
    "eval_loss": 1.382422924041748,
    "eval_rewards/accuracies": 0.5287036895751953,
    "eval_rewards/chosen": -0.7581416964530945,
    "eval_rewards/margins": 0.019789105281233788,
    "eval_rewards/rejected": -0.7779307961463928,
    "eval_runtime": 152.7821,
    "eval_samples": 1080,
    "eval_samples_per_second": 7.069,
    "eval_steps_per_second": 1.767,
    "total_flos": 0.0,
    "train_loss": 1.3895690520604451,
    "train_runtime": 7857.9562,
    "train_samples": 9715,
    "train_samples_per_second": 1.236,
    "train_steps_per_second": 0.01
}